Multiplexed deconvolution of enzyme function in a PLP-dependent protein family

Sequence-based functional annotation of enzymes is an essential step in the discovery and development of new biocatalysts. The vinylglycine ketimine (VGK) subfamily of pyridoxal-phosphate dependent enzymes plays critical roles in sulfur metabolism and is home to a growing range of secondary metabolic enzymes that synthesize noncanonical amino acids. However, discovery of useful new enzymes has been slowed because functional assignments within the VGK sub-family are convoluted by pervasive mis-annotation. Here, we used a whole-cell substrate multiplexed screening approach to rapidly measure catalytic activities of 40 homologs in the VGK subfamily. This strategy gives direct information on enzyme specificity without having to purify each enzyme or measure individual kinetic constants. We identified a thermostable cystathionine γ-lyase from Thermobifida fusca and performed mechanistic and structural studies. For biocatalytic applications, we identified a well-behaved, thermostable, and promiscuous amino acid γ-synthase from Caldicellulosiruptor hydrothermalis (CahyGS). We showed CahyGS can catalyze a stereoselective γ-addition into L-allylglycine, providing preparative-scale access to a unique set of γ-branched amino acids. High resolution crystal structures of CahyGS show an open-closed transition associated with ligand binding and provide a basis for subsequent engineering. Together, these data show how multiplexed screening approaches aid in the rapid deconvolution of enzyme function and identify enzymes with useful properties for enzymology and biocatalysis.